Modeling Cyclic Variability in Integral Compressor Engines for Real-Time Control of NOx Emissions

Abstract

The natural gas industry uses a fleet of integral compressor engines to pressurize the natural gas in pipelines. However, as emissions regulations become increasingly stringent the industry is faced with the undesirable and costly prospect of having to replace their engine fleet. This document describes a three-part research project that is part of an effort to retrofit these engines with real-time emissions control technology that will allow them to remain in compliance with newer emissions regulations.

This research focuses on improving engine diagnostics in order to enhance feedback (or “closed-loop”) control of NOx emissions. It aims to do this primarily by rigorously analyzing many individual cycles, whereas existing techniques typically only analyze an average cycle. Two types of feedback control are considered, referred to in the document as the SwRI method and the Oak Ridge method. These methods are not mutually exclusive and best results can likely be achieved when they are implemented together.

The first task accomplished in this work is the development of a diagnostic cycle simulation capable of analyzing the in-cylinder thermodynamics of many successive individual cycles. Accurately tracking the cycle-to-cycle thermodynamic states of the cylinder gas is key to improving both the SwRI and Oak Ridge methods of control.

The second task accomplished in this work is the development of a method for calculating NOx emissions for many successive individual cycles. This is key to improving the SwRI method, especially as engines emit less and less NOx.

The third task accomplished in this work is the use of the full diagnostic simulation to tune a method for rapidly predicting the residual composition of the gas for a subsequent cycle based on observations of the prior cycle. Such a prediction is vital for the Oak Ridge method to be used with success on the two-stroke engines that constitute the majority of the integral compressor engine fleet.

The completion of this three-part research project revealed that substantial reductions in NOx emissions across many operating conditions could very probably be achieved using the real-time diagnostic and predictive strategies developed for this work. These strategies are mature and ready to be evaluated experimentally on an engine. Additionally, it was discovered that substantial improvements in engine stability and reductions in methane emissions, especially at part-load conditions, could be achieved using the same strategies as for abating NOx emissions.